The origins and evolution of the genus Myosotis L. (Boraginaceae)

The origins and evolution of the genus Myosotis L. (Boraginaceae)

MOLECULAR PHYLOGENETICS AND EVOLUTION Molecular Phylogenetics and Evolution 24 (2002) 180–193 www.academicpress.com The origins and evolution of the ...

530KB Sizes 1 Downloads 41 Views

MOLECULAR PHYLOGENETICS AND EVOLUTION Molecular Phylogenetics and Evolution 24 (2002) 180–193 www.academicpress.com

The origins and evolution of the genus Myosotis L. (Boraginaceae) Richard C. Winkworth,a,b,* J€ urke Grau,c Alastair W. Robertson,b and Peter J. Lockharta,d a

d

Institute of Molecular BioSciences, Massey University, Palmerston North, New Zealand b Institute of Natural Resources, Massey University, Palmerston North, New Zealand c Institut f€ur Systematische Botanik, Universit€at M€unchen., M€unchen, Germany The Allan Wilson Center for Molecular Ecology and Evolution, Massey University, Palmerston North, New Zealand Received 10 August 2001; received in revised form 20 February 2002

Abstract Although morphologically well defined, the phylogeny and taxonomy of Myosotis has been uncertain. In particular it has been unclear whether the genus had a Northern Hemisphere or Australasian origin. However, separate analyses of the ITS and the 30 region of matK, as well as a combined analysis of ITS, 30 matK, the psbA–trnA spacer, and 30 ndhF regions indicate that several distinct lineages exist within Myosotis and strongly support a Northern Hemisphere origin for the genus. Further, the observed transoceanic distributions and levels of genetic divergence between lineages indicate that long distance dispersal has been important for establishing the current geographic range expansion of Myosotis. Our molecular data also suggest that the diversification of Australasian Myosotis has occurred since the late Tertiary and is largely due to radiation within and from New Zealand. This inference is consistent with the findings of recent phylogenetic studies on other New Zealand alpine genera. Our results highlight the important role played by late Tertiary and Quaternary climate change in explaining current floristic diversity. The genetic relationships reported here also suggest that the current infrageneric taxonomy of Myosotis does not fully reflect the evolution of the genus. Ó 2002 Elsevier Science (USA). All rights reserved.

1. Introduction Myosotis consists of about 100 species distributed predominantly in the temperate zones of both hemispheres, with a few taxa occurring in alpine regions of the tropics. Within this broad distribution the genus has two centres of diversity—one in western Eurasia, where approximately 60 taxa occur (Al-Shehbaz, 1991), and the other in New Zealand where approximately 35 species are formally described (Moore, 1961; Moore and Simpson, 1973) and several remain undescribed (Druce, 1993). Outside these two regions the genus is poorly represented with fewer than 10 species restricted to North America, South America, Africa, New Guinea, and Australia (Al-Shehbaz, 1991). Although Myosotis is clearly defined by both vegetative and reproductive morphology the genus is *

Corresponding author. Fax: +203-432-3854. E-mail address: [email protected] (R.C. Winkworth).

considered taxonomically complex, with little consensus on the limits, ranks, and infrageneric classification (Al-Shehbaz, 1991). Earlier taxonomic treatments (e.g., de Candolle, 1846; Stroh, 1941) relied heavily on the nature of the corolla scales and anther exsertion. However, both Moore (1961) and Grau and Leins (1968) have questioned the usefulness of these characters, in particular anther exsertion. The criticism is well illustrated in New Zealand where several species pairs, indistinguishable on the basis of vegetative characteristics, have been assigned to different infrageneric groups due to differences in the degree of anther exsertion (Moore, 1961), Most recently, Grau and co-authors (Grau and Leins, 1968; Grau and Schwab, 1982) have revised the infrageneric taxonomy of Myosotis. Based on their studies of pollen morphology and microscopic characters of the stigma and corolla scales, Grau and Schwab (1982) have proposed that the genus be divided into two sections—section Myosotis and section Exarrhena. The geographic distribution of these morphologically distinct groups

1055-7903/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S 1 0 5 5 - 7 9 0 3 ( 0 2 ) 0 0 2 1 0 - 5

R.C. Winkworth et al. / Molecular Phylogenetics and Evolution 24 (2002) 180–193

181

Table 1 Outline of the infrageneric classification of Myosotis proposed by Grau and Schwab (1982) Subgeneric group

Morphological characteristics

Distribution and content

Section Myosotis

Pollen grains small, constricted at equator and with fine perforations at poles; stigmas two lobed with papillae only slightly differentiated; corolla scales with long papillae Large pollen grains, surface conspicuously sculptured; simple stigmas with large club shaped papillae; corolla scales with short papillae Large pollen grains, but lacking surface sculpturing or perforations found in other groups; form of corolla scales and stigmas as in the austral group

All Eurasian, African, and North American species except those belonging to the discolor group

Section Exarrhena austral group Section Exarrhena discolor group

corresponds to the two centers of Myosotis diversity. Section Myosotis has a Northern Hemisphere distribution, while section Exarrhena occurs in the Southern Hemisphere. The only important exception to this general geographical pattern is a small group of taxa related to the Eurasian annual M. discolor that were included in section Exarrhena because their pollen morphology suggested a close relationship with the Southern Hemisphere taxa (Table 1; Grau and Schwab, 1982). The evolutionary relationships of the infrageneric groupings proposed by Grau and Schwab (1982) have remained unclear. Grau and Leins (1968) suggested that the greater diversity of pollen morphology between austral taxa indicated a Southern Hemisphere origin for the genus. Consistent with this interpretation, both Fleming (1962) and Wardle (1963, 1968) suggested that the ancestors of many New Zealand alpine lineages, including those of Myosotis, were already present in the Southern Hemisphere during the Tertiary period. An alternative hypothesis proposed by Raven (1973) was that the ancestors of the Australasian Myosotis arrived from the Northern Hemisphere only during the late Pliocene or Pleistocene. We determined DNA sequences of the nuclear ITS and three chloroplast DNA markers (the 30 region of the matK gene, the 30 region of the ndhF gene, and the trnK–psbA intergenic spacer) for Myosotis species and representatives from possible outgroup genera within the Boraginaceae. Analyses of these data test hypotheses on the evolution and taxonomy of Myosotis.

2. Materials and methods 2.1. Plant material Tissue samples for this investigation were either silica gel preserved field collections or were obtained from existing herbarium specimens. Appendix A provides details of the accessions used for this investigation.

All Australasian and South American taxa

A small group of taxa distributed in Eurasia, but with one species endemic to east Africa

2.2. DNA extraction, marker amplification, and DNA sequencing Genomic DNA was extracted from dried tissue samples using a cetyltrimethylammonium bromide (CTAB) protocol, modified from Doyle and Doyle (1990). The marker loci were amplified in reaction volumes of 20 lL containing 1 Q solution (Qiagen), 1 PCR buffer (Qiagen), 250 lmol each dNTP (Boehringer Mannheim), 10 pmol each amplification primer, 1 U Taq DNA polymerase (5 U/lL; Qiagen), and 10–100 ng of total cellular DNA. Sequences and combinations of oligonucleotide primers used in PCR amplifications are detailed in Table 2. The reaction conditions for amplification of the matK and ndhF loci were: initial denaturation at 94 °C for 2 min, then 35 cycles of 1 min at 94 °C, 1 min at 50 °C, 2 min at 72 °C with a final extension at 72 °C for 5 min. These reaction conditions were modified for amplification of the remaining loci. For the psbA–trnK intergenic spacer the extension time was reduced to 1 min; for ITS a 1 min extension and 48 °C annealing temperature was used. PCR fragments were then purified using the QIAquick PCR Purification Kit (Qiagen). Both DNA strands of amplification products were characterized using a Perkin–Elmer ABI Prism 377 sequencing protocol. Oligonucleotide primers used for DNA sequencing are detailed in Table 2. For each taxon multiple DNA sequences were aligned by eye and any ambiguity checked against the corresponding electrophoretograms. 2.3. Phylogenetic analyses 2.3.1. Data alignment and tree building Preliminary multiple alignments of DNA sequences were obtained using the progressive alignment procedure implemented in ClustalX (Thompson et al., 1994). These alignments were then inspected visually and checked for misalignment. A conservative approach was taken towards the analysis of the aligned sequences—gapped sequence positions, variable positions immediately flanking gaps, and all ambiguous sequence positions were excluded from tree building analyses. Phylogenetic trees were

182

R.C. Winkworth et al. / Molecular Phylogenetics and Evolution 24 (2002) 180–193

Table 2 Oligonucleotide primers used in the PCR amplification and sequencing of DNA marker loci DNA locus

Primer name

Oligonucleotide sequencea

Used in

matK

trnK3R trnK3AR tK3MY1F tK3MY1FB tK3MY2F

50 50 50 50 50

GATTCGAACCCGGAACTAGTCGG 30 CGTACASTACTTTTGTGTTTMCG 30 CCAATTATGCCAATGATTGCATC 30 CGATACTCTTCTTCCAATTATG 30 CAATCAAAATCTTCTGGAATC 30

Amplification, sequencing Sequencing Sequencing Sequencing (alternative to tK3MY1F) Amplification, sequencing

ndhF

ND1318F ND1656F ND1762R ND2110R

50 50 50 50

GGATTAACYGCATTTTATATGTTTCG 30 ACTTTGTTTGTTGGATGTTTA 30 CCGAAATAAGCTATACTGACT 30 CCCYABATATTTGATACCTTCKCC 30

Amplification, sequencing Sequencing Sequencing Amplification, sequencing

psbA–trnK

trnK3F PSBAR

50 CCGACTAGTTCCGGGTTCGAATC 30 50 CGCGTCTCTCTAAAATTGCAGTCAT 30

Amplification, sequencing Amplification, sequencing

ITS

ITS5 ITS3 ITS2 ITS4

50 50 50 50

Amplification, sequencing Sequencing Sequencing Amplification, sequencing

GGAAGTAAAAGTCGTAACAAGG 30 GCATCGATGAAGAACGTAGC 30 GCTACGTTCTTCATCGATGC 30 TCCTCCGCTTATTGATATGC 30

a With the exception of the ITS primers (from White et al., 1990) and the ndhF amplification primers (ND1318F and ND2110R from Olmstead and Sweere, 1994) the oligonucleotide primers were designed during this study.

constructed for Myosotis with parsimony and quartet puzzling, the latter using a maximum likelihood (ML) optimality criterion, as implemented in PAUP*4.0b8 (Swofford, 2001). Parsimony analyses were performed using the heuristic search option and the ‘‘tree-bisectionreconnection’’ (TBR) swapping algorithm with accelerated transformation (ACCTRAN) optimization used to infer branch (edge) lengths. For quartet puzzling with ML, a HKY 85 invariable sites model (Swofford et al., 1996) was used in all analyses. Empirical base frequencies were assumed and other model parameters (e.g., the transition/tranversion ratio and proportions of invariable sites) were estimated on Neighbor Joining (NJ) trees. 2.3.2. Rooting the phylogeny Evolutionary relationships within the Boraginaceae are not well understood (Al-Shehbaz, 1991). The uncertain taxonomy and current lack of a thoroughly sampled molecular phylogeny for the Boraginaceae made it unfeasible to a priori identify appropriate outgroups for Myosotis. For this reason, species from several genera of the Boraginaceae—Borago, Echium, Eritrichium, Myosotidium, Plagiobothrys, and Symphytum were sequenced (Appendix A) and phylogenetic analyses of these data used to infer the position of the root in ITS and matK phylogenies for Myosotis. For each potential outgroup all possible placements of the DNA sequence on to the most parsimonious unrooted Myosotis tree were considered. Pairwise comparisons of trees with different outgroup placements were then evaluated for significance using Kishino–Hasegawa (K–H; Kishino and Hasegawa, 1989) and Shimodaira–Hasegawa (S–H; Shimodaira and Hasegawa, 1999) sites tests as implemented in PAUP*4.0b8. Strictly, the K–H sites test compares topologies for two trees chosen without reference to the sequence data. In contrast, the more

conservative S–H sites test compares two trees belonging to a set of possible trees that may have been chosen with reference to the sequence data (Goldman et al., 2000). The placements of two potential outgroup taxa (Echium vulgare and Plagiobothrys albiflorus; see Appendix A) were also evaluated on a combined data set that included sequences from the ITS, matK, ndhF, and psbA–trnK regions for a representative sample of Myosotis. A preliminary phylogenetic analysis of these data appeared in Winkworth et al. (1999) and we expand on those here. Combining sequence data from different gene loci have been shown to improve phylogenetic resolution in some situations (Cunningham, 1997). However, the approach can be inappropriate when a common mechanism of sequence evolution between loci cannot be assumed and when rates of change are significantly different between data sets (Huelsenbeck et al., 1996; Lockhart and Cameron, 2001). Prior to testing outgroup placements with the combined data we tested for partition incongruence using the partition homogeneity test (Michevich and Farris, 1981) as implemented in PAUP*4.0b8. 2.3.3. Age of the austral lineage The length of time that the austral lineage has been in the Southern Hemisphere was investigated using the simplified phylogenetic scheme shown in Fig. 1. An upper bound for this age was assumed to correspond to the distance between ancestral node III and extant taxa within the Southern Hemisphere group. For our test, we built branch and bound ML trees (details of the model were the same as for quartet puzzling analyses). These contained the nine Eurasian taxa, inferred from the analyses shown in Figs. 2 and 3 to be most closely related to the Southern Hemisphere austral group. They also contained either M. albo-sericea or

R.C. Winkworth et al. / Molecular Phylogenetics and Evolution 24 (2002) 180–193

183

In our analyses, we excluded sequences which did not meet requirements of rate constancy as inferred using a relative rates test (Steel et al., 1996) implemented in SplitsTree 3.1 (Huson, 1998).

3. Results 3.1. Aligned DNA sequences

Fig. 1. Estimation of the possible age of the Southern Hemisphere universal common ancestor. Grey shading indicates extant taxa (circles) or groups of taxa (triangle) and open circles represent inferred ancestral taxa. Ancestors I and II are assumed to have had a Eurasian distribution. Ancestor III may have been present in either Eurasia or the Southern Hemisphere, while IV was the most recent common ancestor of the extant austral radiation. For estimates of the maximum possible age for the austral ancestor we assume the most parsimonious reconstruction—III had a Northern Hemisphere distribution and IV was Australasian.

M. exarrhena, both from the austral group. The topology of the ML tree and optimal branch lengths were calculated and constrained to these values, except for the length of the edge that led to the austral representative. The length of this branch was varied such that it corresponded to a calibrated time of between 0 and 100 million years (MY). We used several rate calibrations for sequence divergence and time. The first made use of the New Zealand fossil pollen record and observed sequence diversity among the extant austral species shown in Fig. 3. That is, we assumed that the first appearance of Myosotis pollen in the New Zealand fossil record corresponded to the immediate ancestor (node IV in Fig. 1) of the extant radiation of Australasian species, and not the last common ancestor (node III in Fig. 1) of Southern Hemisphere and Eurasian species. The average genetic distance (path length) between Southern Hemisphere species was calculated for independent paths that crossed the node that corresponded to node IV in Fig. 1. This genetic distance was calibrated with the earliest record of Myosotis pollen in New Zealand (2 MY; Mildenhall, 1981). A second calibration of sequence divergence with absolute time was made by comparison of ITS sequences from Myosotis with those from the Juan Fernandez Island endemic genus Dendroseris (Asteraceae). In an earlier study, Sang et al. (1994) reported pairwise comparisons for ITS sequences from all Dendroseris taxa, and by assuming an age of 4 MY for the Juan Fernandez Islands, estimated an evolutionary rate of 3:94  103 substitutions/site/MY for ITS. Since their approach used nonindependent paths between taxa we also re-calculated the substitution rate for Dendroseris ITS sequences using independent paths from their data set.

The high degree of similarity between the DNA sequences characterized for each molecular marker resulted in data matrices that contained little alignment ambiguity. Statistics from the aligned data matrices for Myosotis sequences are presented in Table 3. 3.2. Phylogenetic gene trees Nuclear ITS and chloroplast matK regions were analyzed under two evolutionary tree building methods. Under maximum parsimony a single most parsimonious tree of 75 steps was recovered in analyses of the matK data set; this tree had a consistency index (CI) of 0.933 and retention index (RI) of 0.956. ITS sequences produced three equally parsimonious trees of 194 steps with CI of 0.691 and RI of 0.855. These trees differed only in the relationships inferred between Myosotis alpestris, M. lithospermifolia, and M. semiamplexicaulis. Fig. 2 presents 50% majority rule consensus trees for both the ITS and matK data sets. These data were also analyzed using quartet puzzling with a ML optimality criterion; trees from these analyses are shown in Fig. 3. Note that although indels were removed from analyses, only one was incompatible with the phylogenetic reconstructions based on nucleotide substitutions. That is, only a single 1 bp indel in the ITS region shared by the M. sylvatica lineage and the North American species, M. macrosperma, and M. verna (sequence position 432) was not compatible with the evolutionary trees reported. The topology of trees recovered with chloroplast and nuclear markers were highly congruent with each other. Congruence between trees built from these data sets is highlighted by the observation that sequences from only two taxa differ in their placement. In the ITS gene tree, the sequence from M. persoonii is closest to species from Southern Hemisphere, whereas in matK trees M. persoonii groups closely with several Eurasian taxa (compare Figs. 2 and 3A with 2 and 3B). The matK sequence for the African species M. abysinnica groups it with members of the ‘‘discolor’’ species group, however the ITS sequence places this species with representatives of section Myosotis. The gene trees presented here suggest five distinct phylogenetic groups terminal to branches A–E in Figs. 2 and 3. Group A contains several Eurasian representatives and all Southern Hemisphere species. The

184

R.C. Winkworth et al. / Molecular Phylogenetics and Evolution 24 (2002) 180–193

remaining four groups have species restricted to the Northern Hemisphere, with the exception of two southern African species that have been derived from two of these. Although our analyses indicate that genetic groups within Myosotis are distinct from each other, relationships between them are not resolved as bifurcating. Rather lineages are mostly separated from each other by polytomies or by weak bifurcations with low support values (e.g., Figs. 2 and 3).

support under QP). Support for this inferred relationship is not significant under pairwise sites tests. However, there is less support for competing hypotheses. These include hypotheses that taxa from other Southern Hemisphere lands are more basal to the New Zealand species in outgroup rooted phylogenies (less than 1% support under QP; Fig. 4). As will be discussed, these results are consistent with New Zealand as the most likely center for the origin of the extant Southern Hemisphere species.

3.3. Relative genetic diversity and outgroup rooting

3.5. Age of the Australasian lineage in the Southern Hemisphere

The branch lengths in gene trees derived from matK and ITS sequences (Figs. 2 and 3) indicate that the greatest genetic diversity in Myosotis occurs amongst the Northern Hemisphere lineages—an observation consistent with a Northern Hemisphere origin for the genus. This hypothesis was tested by investigating the preferred placement of six potential outgroup sequences onto the optimal matK and ITS phylogenies for Myosotis. The sequences from these six taxa were found to be genetically divergent from their orthologs in Myosotis species, consistent with these species being outgroups to the genus Myosotis. In the tests made, hypotheses proposing a Northern Hemisphere origin for the genus (trees in which the branch to the outgroup joined an edge within a Eurasian lineage) were always favored over those suggesting that Myosotis arose in the Southern Hemisphere (trees in which the branch to the outgroup either joined within the austral radiation or the branch subtending this group). Arrows on Fig. 2 indicate the optimal rooting position for each of the outgroups. While a Northern Hemisphere origin is always favored, gene trees supporting a Southern Hemisphere origin were only significantly worse in 44% of the Kishino–Hasegawa tests at the P < 0:05 level, and were less significant under the more conservative Shimodaira–Hasegawa sites test. To investigate whether significance could be obtained by using longer sequences, a combined data set comprising sequences from ITS, matK, ndhF, and trnK–psbA loci were jointly analyzed. These additional sequences were determined for the outgroup taxa Echium vulgare and Plagiobothrys albiflorus as well as a representative sample of Myosotis species. Again the optimal placement of the outgroup sequences were on edges within Northern Hemisphere lineages. More importantly, under Kishino–Hasegawa and Shimodaira–Hasegawa sites tests, all but one of the outgroup placements (i.e., 95.5% of tests) suggesting a Southern Hemisphere origin were significantly worse at the P < 0:05 level. 3.4. Dispersal in the Southern Hemisphere Quartet puzzle and maximum parsimony trees for the ITS region (Fig. 2A, 3A, 4) all suggest New Zealand taxa are basal to the other species of the austral group (93%

The length of time that the austral lineage has been present in the Southern Hemisphere was investigated by examining sequence divergence and various rate calibrations for sequence diversity and absolute time. The first and most conservative calibration was derived from the earliest fossil record for Myosotis in New Zealand. This was 1:1  103 –3:4  103 substitutions/site/MY, and using these estimates our ML inference suggests that the earliest possible Southern Hemisphere ancestor existed 2.0–14.7 MYA (Table 4). A second rate calibration, derived from a study of Dendroseris ITS sequence diversity, suggested ITS sequence evolution at 3:3  103 –5:8  103 substitutions/site/MY (a range which included the point estimate of Sang et al., 1994). With this calibration method the austral Myosotis lineage is estimated to have existed in the Southern Hemisphere 1.2–4.9 MYA (Table 4). The variance on these estimates was examined by comparing log-likelihood scores for trees constrained for topology and branch lengths. When M. albo-sericea was used as a representative of the austral group, trees suggesting the presence of a Southern Hemisphere lineage older than 16.4 MY (assuming the conservative pollen calibration) and 5.5 MY (assuming the conservative Dendroseris ITS calibration) were rejected (P < 0:05). When M. exarrhena was used as the Southern Hemisphere representative, the age of its lineage in the Southern Hemisphere was found to be less than 29.5 and 9.8 MY, respectively (P < 0:05). These results reject the possibility that the austral group has a relictual Gondwanan distribution.

4. Discussion 4.1. Myosotis phylogeny: the similarity between gene trees Few incompatibilities exist between trees derived from the ITS and matK loci. This high degree of congruence between gene trees derived from different markers suggests that the inferred phylogenetic patterns reflect a common evolutionary history for Myosotis.

R.C. Winkworth et al. / Molecular Phylogenetics and Evolution 24 (2002) 180–193

185

Fig. 2. Bootstrap majority rule consensus trees reconstructed using a parsimony criterion; branch lengths have been estimated under ACCTRAN (using PAUP 4.0b8). (A) Data set for 34 Myosotis taxa from the ITS region of the nrDNA (610 nucleotides). (B) Data set for 34 Myosotis taxa from the 30 region of the chloroplast matK gene locus (867 nucleotides). Main genetic groupings have been indicated by a circled capital letter on the branch subtending each lineage. Subgeneric groups of Grau and Schwab (1982) are also indicated by different shadings in the rectangular box to the right of the figure—section Exarrhena austral group are white; section Exarrhena discolor group are light grey; section Myosotis are shown as dark grey. Preferred outgroup rooting positions are marked by arrows, outgroups are denoted by two letter code—Bo., Borago officinalis; Ec., Echium vulgare; Er., Eritrichium nanum; Ho., Myosotidium hortensia; Pl., Plagiobothrys albiflorum; Sy., Symphytum  uplandicum. Generalized distributions for taxa are indicated by abbreviations: Afr., Africa; Aus., mainland Australia; Eur., Europe; Mid., Middle East; NAm., North America; NZ, New Zealand; SAm., South America; Sub., Subantarctic Islands; Tas., Tasmania. Bootstrap values (250 replicates) for internal branches are given.

Differences between gene trees included the placement of M. persoonii and M. abysinnica (Figs. 3A and B). These incongruences are possibly explained by either homoplasy or ‘‘chloroplast capture,’’ since sequence analysis of independent PCR amplified products gave identical results. Characterization of further chloroplast and nuclear markers for populations of M. persoonii and M. abysinnica could help to distinguish between these possible explanations. 4.2. Origins and early evolution of Myosotis Based on the greater diversity of pollen morphology in Australasian Myosotis, Grau and Leins (1968)

proposed that the Eurasian lineages arose from taxa with long histories in the Southern Hemisphere. However, the greater genetic diversity between Northern Hemisphere lineages, relative to that observed among austral representatives, suggests that the genus arose in the Northern Hemisphere. Tests of outgroup placement on to Myosotis phylogenies support a Eurasian origin. In particular, analyses using a combined data set favored a Northern Hemisphere ancestry. Whilst our molecular results strongly suggest that Myosotis arose in the Northern Hemisphere it was not possible to identify the most ancestral lineage as no single optimal root position was identified.

186

R.C. Winkworth et al. / Molecular Phylogenetics and Evolution 24 (2002) 180–193

Fig. 2. (continued).

Our molecular analyses identified five distinct lineages within Myosotis; molecular characterization of additional species may reveal others. Although these major groups are genetically well differentiated, the relationships between them are poorly resolved (i.e., related to each other either by polytomies or short branches with low support values) suggesting that they may have originated during a period of rapid diversification, perhaps shortly after the genus arose. 4.3. Dispersal and the worldwide distribution of Myosotis An important event in the evolution of Myosotis was the dispersal and establishment of the genus in Australasia. Consistent with the hypothesis of Raven (1973), our phylogenetic analyses suggest that a single, mid, or late Tertiary (<20 MYA) transoceanic dispersal event by a Northern Hemisphere ancestor established Myosotis in New Zealand. In ITS gene trees, the discolor group

species M. persoonii is the closest Eurasian relative to austral species; a finding consistent with Grau and Schwab’s (1982) suggestion that the discolor group is closely related to the austral species. The chloroplast data do not provide sufficient resolution to address this issue. The extent of sequence divergence within the austral group suggests that Myosotis may have been in the Southern Hemisphere longer than is indicated by fossil deposits in New Zealand. In this case, species extinctions associated with Quaternary climate change cannot be discounted and consideration of this possibility may also have important implications for understanding plant dispersal routes into New Zealand. Raven (1973) suggested that dispersal of many alpine groups, including that of Myosotis, to New Zealand occurred via the Australian and New Guinean Alps. Our analyses do not support this hypothesis. However, if the isolated alpine zones of Australia and New Guinea were seriously restricted as climate change

R.C. Winkworth et al. / Molecular Phylogenetics and Evolution 24 (2002) 180–193

187

Fig. 3. Quartet puzzle trees with branch lengths estimated using ML (using PAUP 4.0b8). (A) Data set for 34 Myosotis taxa from the ITS region of the nrDNA (610 nucleotides; pinv ¼ 0:803 and ti=tv ¼ 1:792). (B) Data set for 34 Myosotis taxa from the 30 region of the chloroplast matK gene locus (867 nucleotides; pinv ¼ 0:639 and ti=tv ¼ 1:412). Puzzle support values for internal branches are given. Genetic lineages, taxonomic groups, and the distribution of M. australis accessions are denoted as for Fig. 2.

during the late Tertiary/Quaternary as in New Zealand (Lee et al., 2001), such ‘‘stepping stones’’ may no longer contain the ancestral austral species. If so, then it may be problematic to test dispersal routes into New Zealand for plant groups using analyses of extant species alone. Despite the concern that sequence analyses may be unable to pinpoint dispersal routes into New Zealand, our results do indicate transoceanic dispersals from New Zealand to other Southern Hemisphere lands (Fig. 4). The phylogenetic inference that some dispersal events have occurred in northerly and westerly directions is of particular interest. Previous authors have suggested the importance of west wind drift, the dominant meteorological pattern at high southern latitudes, in facilitating plant dispersal in an easterly direction (e.g., Fleming, 1979; McPhail, 1997; Mildenhall, 1981; Pole, 1994; Raven, 1973). While our results provide some evidence for the possibility of dispersal facilitated by west wind drift (i.e., from New Zealand to South America for M. albiflora), they also suggest that successful dispersal events have occurred to the north and west (i.e., from New Zealand to Australia for M. australis; and from New Zealand to New Guinea also for M. australis), movements that are counter to the direction of west wind drift.

Other instances of transoceanic dispersal by representatives of Eurasian lineages are also suggested by our phylogenetic graphs. Specifically analyses of both nuclear and chloroplast markers suggest that the establishment of Myosotis in Africa and North America has been the result of dispersal from Eurasian sources. Few data are available on the dispersal mechanisms in Myosotis (e.g., Bresinsky, 1963; Ridley, 1930; Steyermark, 1963). Given that Myosotis nutlets appear to be poorly adapted for wind dispersal, this mechanism seems unlikely. Rather the trichomes found on the calyxes and pedicels of many Myosotis species (Al-Shehbaz, 1991; Heywood, 1993) suggest that animal vectors may facilitate dispersal. Indeed, bird-mediated dispersal has been favored as a mechanism for transoceanic dispersal of disseminules in several plant groups in the South Pacific (Carlquist, 1996; Godley, 1967; Swenson and Bremer, 1997). 4.4. Recent species diversification In direct contrast to the reported morphological diversity of austral Myosotis species (Grau and Leins, 1968; Grau and Schwab, 1982; Winkworth et al., 1999), our molecular investigations show that this group is

188

R.C. Winkworth et al. / Molecular Phylogenetics and Evolution 24 (2002) 180–193

Fig. 3. (continued).

Table 3 Statistics from the aligned data matrices of Myosotis DNA sequences

No. of taxa sequenced Sequence length range (bp) Aligned length (bp) No. of indels No. of excluded sites No. of constant sites No. of varied sites No. of parsimony informative sites % GC content range (all sites included) % GC content mean (all sites included) % GC content range (varied sites only) % GC content mean (varied sites only)

ITS

Chloroplast matK

ndhF

trnK–psbA

34 646–649 659 11 49 511 99 68 55.6–58.9 57.2 42.4–58.6 52.5

34 882–923 925 11 58 801 66 34 32.4–33.4 33.0 51.5–65.7 63.3

14 725–731 731 1 6 687 36 19 27.1–28.1 27.4 44.4–63.4 49.4

14 223–250 250 6 28 211 11 4 32.9–33.3 32.9 36.4–54.5 45.5

largely undifferentiated at several presumably neutrally evolving DNA loci. This pattern of low genetic and high morphological diversity suggests that Myosotis has radiated only recently in the Australasian region—an interpretation supported by molecular clock analyses on ITS sequences which suggest that the austral Myosotis species have radiated only since the Pliocene (Table 4). The ITS gene trees also suggest that considerable mor-

phological diversification associated with this radiation has occurred in New Zealand (Fig. 4). The diversification of Myosotis in New Zealand correlates with the late Tertiary and Quaternary period—a time of considerable climatic fluctuation and geological upheaval (Raven, 1973; Wardle, 1968), Further, the branching patterns within the major Northern Hemisphere lineages may suggest that environmental changes have also been

R.C. Winkworth et al. / Molecular Phylogenetics and Evolution 24 (2002) 180–193

189

Fig. 4. Quartet-puzzle tree with branch lengths estimated using ML (PAUP 4.0b8), made using a dala set of 634 nucleotides from the ITS region of the nrDNA for 22 Myosotis taxa ðpinv ¼ 0:803 and ti=tv ¼ 1:792). Puzzle support values for internal branches are given. Subgeneric groups and the distribution of M. australis accessions as for Fig. 2.

Table 4 Age estimate for the origins of the austral lineage of Myosotis Age estimate for

M. albo-sericea M.exarrhena

Divergence in optimal ML tree

0.00639 0.01563

important for establishing the extant species diversity of Eurasian Myosotis. Our molecular evidence for recent diversification in Myosotis is comparable with similar results reported for other New Zealand plant groups (e.g., see discussion in Winkworth et al., 1999) and from genetic studies on continental floras of Europe and North America. These investigations suggest that environmental change during the late Tertiary and Quaternary periods, both at global and local scales, is of great importance in understanding the development of the world’s modern floristic biodiversity (Comes and Kadereit, 1998). 4.5. Implications of molecular analyses for the intrageneric classification of Myosotis The issue of monophyly in taxonomy is well debated and it is widely accepted that only monophyletic groups should be formally recognized in taxonomic classification (Donoghue and Cantino, 1988). While the

Estimated ages (MY) Re-estimated Dendroseris rate

Pollen based calibration

1.2–2.0 2.8–4.9

2.0–6.1 4.8–14.7

molecular results reported here support certain aspects of the intrageneric classification proposed by Grau and Schwab (1982), in particular the close relationship between the discolor and austral groups of sect. Exarrhena, they do not provide evidence for the monophyly of the currently recognized sections. Our molecular results may indicate the need for revised intrageneric classification.

Acknowledgments We thank Tristan Armstrong, Andrea Brandon, Fritz Ehrendorfer, Mark Large, David Havell, Jo Ward, Brian, and Chris Rance as well as staff at the listed herbaria and gardens (Appendix A) for assistance with tissue samples. This work was supported by funding from the New Zealand Marsden Foundation, Massey University and the Alexander von Humboldt Foundation.

190

R.C. Winkworth et al. / Molecular Phylogenetics and Evolution 24 (2002) 180–193

Appendix A. Details of accessions of Myosotis and outgroup taxa Taxon

Location

Herbarium referencea

GenBank Accession

Borago officinalis L.

Palmerston North, North Island, New Zealand (naturalized introduction) Lake Ohau, South Island, New Zealand (naturalized introduction) Switzerland

MPN 24675

ITS: AY092898 matK: AY092892

MPN 24676

ITS: AY092900 matK: AY092893 ndhF: AY092890 psbA-trnK: AY092952 ITS: AY092901 matK: AY092894 ITS: AY092902 matK: AY092895

Echium vulgare L.

Eritrichium nanum (L.) Schrader ex Gaudin Myosotidium hortensia (Decne) Baill. Myosotis abyssincia Boiss. & Reuter Myosotis afropalustris C.H.Wr.

Palmerston North, North Island, New Zealand Ethiopia Natal, South Africa

Hertel 25764b No voucher (nursery origin) De Wilde 6944b

Myosotis albiflora Banks & Sol.

Punta Arenas, Chile

K. Balkwill & M.J. Balkwill 5260b MPN 24679

Myosotis albo-sericea Hook. F.

Leaning Rock, South Island, New Zealand

Site voucher CHR 416092c

Myosotis alpestris F.W. Schmidt

Karten, Hochobir, Austria

‘‘FE-M2’’

Myosotis arvensis (L.) Hill

Bavaria, Germany

H. Gr€ oger 1018b

Myosotis australis R. Br.

Tasmania, Australia

MPN 24677

Myosotis australis R. Br.

MPN 24678

Myosotis australis R. Br. (identified as M. saruwagedica) Myosotis australis R. Br. ‘‘Yellow’’ Myosotis brockiei L. Moore et M. Simpson Myosotis cadmea Boiss.

Mt. Kozciuscko, Australia New Guinea

CHR 198545

Mt. Ben More, South Island, New Zealand Cobb Gorge, South Island, New Zealand Greece

Site voucher CHR 365380c Site voucher CHR 311717c Stainton 7320b

Myosotis capitata Hook. F.

Cultivated (B. Rance)

MPN 24680

ITS: AY092904 matK: AY092856 matK: AY092857 ITS: AY092906 matK: AY092858 ndhF: AY092848 psbA-trnK: AY092938 ITS: AY092905 matK: AY092859 ndhF: AY092842 psbA-trnK: AY092939 ITS: AY092907 matK: AY092860 ndhF: AY092854 psbA-trnK: AY092951 ITS: AY092908 matK: AY092861 ndhF: AY092843 psbA-trnK: AY092949 ITS: AY092910 matK: AY092863 ndhF: AY092844 psbA-trnK: AY092944 ITS: AY092909 matK: AY092862 ITS: AY092933 matK: AY092884 ITS: AY092911 ITS: AY092912 ITS: AY092913 matK: AY092864 ITS: AY092915 matK: AY09B866 ndhF: AY092853 psbA-trnK: AY092943

R.C. Winkworth et al. / Molecular Phylogenetics and Evolution 24 (2002) 180–193

191

Appendix A. (continued) Taxon

Location

Herbarium referencea b

Myosotis congesta Shuttlew. Ex Alb. & Reynier Myosotis debilis Pomel

Greece

Phitos M-33

Spain

Gomez Vigide317b

Myosotis decumbens Host. ssp. decumbens Myosotis discolor Pers.

Alpi Maritime, Sch€ onswetter & Tribsch Kahuterawa Valley, North Island, New Zealand (naturalized introduction) South-eastern NSW, Australia Broken River, South Island, New Zealand

‘‘FE-M’’

Myosotis exarrhena (R. Br.) F. Mueller Myosotis goyenii Petrie

MPN 11910

CBG 9519354 Site voucher CHR 278306c

Myosotis incrassata Guss.

Greece

Merxtm€ uller & Wiedmann 20130b MPN 24681

Myosotis laxa Lehm. ssp. caespitosa (C.F Schultz) Hyl.

Hopkins River, South Island, New Zealand

Myosotis lithospermifolia Hornem.

Persia

Rechinger 6521b

Myosotis macrantha (Hook, F.) Benth. Et Hook. F.

Hooker Valley, South Island, New Zealand

Site voucher CHR 252807c

Myosotis macrosperma Engelm.

USA

MPN 24682

Myosotis matthewsii L. Moore

Cultivated (Percy’s Reserve, Wellington, New Zealand)

No voucher

Myosotis persoonii Rouy

Spain

Zubizarreta 5327b

Myosotis propinqua Fisch. & Mey.

Persia

Rechinger 39832b

Myosotis rakiura L. Moore

Curio Bay, South Island, New Zealand

MPN 24686

Myosotis refracta Boiss. ssp. refracta Greece

Gr€ oger 1463ab

Myosotis rehsteineri Wartm.

Bavaria, Germany

D€ orrb

Myosotis ruscinonensis Rouy

France

Myosotis semiamplexicaulis DC.

South Africa

Kunz & Reichstein M-310b Acocks 21310b

GenBank Accession ITS: AY092916 matK: AY092867 ITS: AY092917 matK: AY092868 ITS: AY092918 matK: AY092869 ITS: AY092919 matK: AY092870 ndhF: AY092852 psbA-trnK: AY092948 ITS: AY092920 matK: AY092871 ITS: AY092921 matK: AY092872 ndhF: AY092846 psbA-trnK: AY092942 ITS: AY092922 matK: AY092873 ITS: AY092914 matK: AY092865 ndhF: AY092851 psbA-trnK: AY092946 ITS: AY092923 matK: AY092874 ITS: AY092924 matK: AY092875 ndhF: AY092849 psbA-trnK: AY092940 ITS: AY092925 matK: AY092876 ndhF: AY092855 psbA-trnK: AY092947 ITS: AY092926 matK: AY092877 ndhF: AY092847 psbA-trnK: AY092941 ITS: AY092927 matK: AY092878 ITS: AY092928 matK: AY092879 ITS: AY092929 matK: AY092880 ndhF: AY092845 psbA-trnK: AY092945 ITS: AY092930 matK: AYQ92881 ITS: AY092931 matK: AY092882 ITS: AY092932 matK: AY092883 ITS: AY092934 matK: AY092885

192

R.C. Winkworth et al. / Molecular Phylogenetics and Evolution 24 (2002) 180–193

Appendix A. (continued) Taxon Myosotis stricta Link Myosotis sylvatica Ehr. ex Hoffm. ssp. Sylvatica

Herbarium referencea

Location

Myosotis verna Nutt.

Bavaria, Germany Palmerston North, North Island New Zealand (naturalized introduction), Hopkins River, South Island, New Zealand USA

Myosotis vestergrenii Stroh

Ethiopia

MO 3808817

Plagiobothrys albiflorus (Griseb.) R.L. Perez-Mor.

Paso Cardenal Samore, Argentina

MPN 24689

Symphytum  uplandicum Nyman

Wellington, North Island, New Zealand (naturalized introduction)

MPN 24690

Myosotis uniflora Hook. f.

F€ orther 7895 MPN 24687

b

GenBank Accession matK: AY092886 ITS: AY092935 matK: AY092887 ndhF: AY092850 psbA-trnK: AY092950

MPN 24688 Taylor 3489b

ITS: AY092936 matK: AY092888 ITS: AY092937 matK: AY092889 ITS: AY092899 matK: AY092896 ndhF: AY092891 psbA-trnK: AY092953 ITS: AY092903 matK: AY092897

a

Herbarium codes. CBG—Australian National Botanic Gardens, Canberra, ACT, Australia; CHR—Landcare Research New Zealand, Christchurch, New Zealand; MO—Missouri Botanical Garden, Saint Louis, Missouri, USA; MPN—Massey University, Palmerston North, New Zealand. b Herbanum accession lodged at the Staatsherbarium M€ unchen [MSB]. c Due to the small size of these populations or rarity of the species in general no voucher was taken for this study. However, these populations are represented by herbarium collections made by previous workers.

References Al-Shehbaz, I.A., 1991. The genera of Boraginaceae in the Southeastern United States. J. Arnold Arboretum, Suppl. Ser. 1, 1–169. Bresinsky, A., 1963. Bau, Entwicklungsgeschicte und Inhaltsstoffe der Elaisomen. Bibliot. Bot. 126, 54 pp.. de Candolle, A.P., 1846. Boragineae. Prodromos 10, 105–116. Carlquist, S., 1996. Plant dispersal and the origin of Pacific island floras. In: Keast, A., Miller, S.E. (Eds.), The Origin and Evolution of Pacific Island Biotas, New Guinea to Eastern Polynesia: Patterns and Processes. SPB Academic Publishing, pp. 153–164. Comes, H.P., Kadereit, J.W., 1998. The effects of Quaternary climatic changes on plant distribution and evolution. Trends Plant Sci. 3, 432–438. Cunningham, C.W., 1997. Can three incongruence tests predict when data should be combined? Mol. Biol. Evol. 14, 733–740. Donoghue, M.J., Cantino, P.D., 1988. Paraphyly, ancestors, and the goals of taxonomy: a botanical defence of cladism. Botanical Rev. 54, 107–128. Doyle, J.J., Doyle, J.L., 1990. Isolation of plant DNA from fresh tissue. Focus 12, 13–15. Druce, A.P., 1993. Indigenous higher plants of New Zealand, Unpublished checklist, Landcare Research, Lower Hutt. Fleming, C.A., 1962. New Zealand biogeography—a paleontologists approach. Tuatara 10, 53–104. Fleming, C.A., 1979. The geological history of New Zealand and its life. Auckland University Press, New Zealand. Godley, E.J., 1967. Widely distributed species, land bridges, and continental drift. Nature 214, 74–75. Goldman, N., Anderson, J.P., Rodrigo, A.G., 2000. Likelihoodbased tests of topologies in phylogenetics. Syst. Biol. 49, 652– 670.

Grau, J., Leins, P., 1968. Pollenkorntypen und Sektionsgliderung der Gattung Myosotis L. Berichte der Dt. Bot. Gesellschaft 81, 107–117. Grau, J., Schwab, A., 1982. Mikromerkmale der bl€ ute zur gliederung der Gattung Myosotis. Mitt. Bot. M€ unchen 18, 9–58. Heywood, V., 1993. Flowering plants of the world. B T Batsford Ltd, London. Huelsenbeck, J.R., Bull, J.J., Cunningham, C.W., 1996. Combining data in phylogenetic analysis. Trends Ecol. Evol. 11, 152–157. Huson, D.H., 1998. SplitsTree: analysing and visualising evolutionary data. Bioinformatics 14, 68–73. Lee, D.E., Lee, W.G., Mortimer, N., 2001. Where and why have all the flowers gone? Depletion and turnover in the New Zealand Cenozoic angiosperm flora in relation to palaeogeography and climate. Aust. J. Botany 49, 341–356. Lockhart, P.J., Cameron, S.A., 2001. Trees for bees. Trends Ecol. Evol. 16, 84–88. Kishino, H., Hasegawa, M., 1989. Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order of Hominiodea. J. Mol. Evol. 29, 170–179. Mildenhall, D.C., 1981. New Zealand late cretaceous and cenozoic plant biogeography: a contribution. Paleogeo., Paleoclim., Paleoecol. 31, 197–233. Michevich, M.F., Farris, J.S., 1981. The implications of congruence in Menidia. Syst. Zool. 30, 351–370. McPhail, M.K., 1997. Comment on M. Pole (1994) The New Zealand flora-entirely long distance dispersal? J. Biogeogr. 21, 625–635, J. Biogeogr. 24, 113–117. Moore, L.B., 1961. Boraginaceae. In: Allan, H.H. (Ed.), Flora of New Zealand, vol. I. Government Printer, Wellington, pp. 806–833. Moore, L.B., Simpson, M., 1973. A new Myosotis from north-west Nelson. N. Z. J. Botany 11, 163–170.

R.C. Winkworth et al. / Molecular Phylogenetics and Evolution 24 (2002) 180–193 Olmstead, R.G., Sweere, J.A., 1994. Combining data in phylogenetic systematics: an empirical approach using three molecular data set in the Solanaceae. Syst. Biol. 43, 467–481. Pole, M., 1994. The New Zealand flora-entirely long distance dispersal. J. Biogeogr. 21, 625–635. Raven, P.H., 1973. Evolution of the subalpine and alpine plant groups in New Zealand. N. Z. J. Bot. 11, 177–200. Ridley, H.N., 1930. The dispersal of plants throughout the world. Kent, England. Sang, T., Crawford, D.J., Kim, S.-C., Stuessy, T.F., 1994. Radiation of the endemic genus Dendroseris (Asteraccae) on the Juan Fernandez Islands: evidence from sequences of the ITS region of the nuclear ribosomal DNA. Am. J. Bot. 81, 1494–1501. Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of loglikelihoods with applications to phylogenetic inference. Mol. Biol. Evol. 16, 1114–1116. Steyermark, J.A., 1963. Flora of Missouri. Ames, Iowa. Stroh, G., 1941. Die Gattung Myosotis L. Versuch einer system€ bersicht u €ber die Arten. Beih. Bot. Centralbl. 61, 317– atischen U 345. Swenson, U., Bremer, K., 1997. Patterns of floral evolution of four Asteraceae genera (Senecioneae, Blennospermatinae) and the

193

origin of white flowers in New Zealand. Syst. Biol. 46, 407– 425. Swofford, D.L., 2001. PAUP*. Phylogenetic Analysis Using Parsimony (* and Other Methods).Version 4. Sinauer Associates, Sunderland, MA. Thompson, J.D., Higgins, D.G., Gibson, T.J., 1994. CLUSTAL W: improving the sensitivity of progressive sequence alignment through sequence weighting, position specific gap penalties and weight matrix choice. Nucleic Acids Res. 22, 4673–4680. Wardle, P., 1963. Evolution and distribution of the New Zealand flora, as affected by Quaternary climates. N. Z. J. Bot. 1, 3–17. Wardle, P., 1968. Evidence for an indigenous pre-quaternary element in the mountain flora of New Zealand. N. Z. J. Bot. 6, 120– 125. White, T.J., Bruns, T., Lee, S., Taylor, J., 1990. Amplification and direct sequencing of fungal ribosomal genes for phylogenetics. In: Innis, M., Gelfand, D., Sninsky, J., White, T. (Eds.), PCR Protocols: A Guide to Methods and Applications. Academic Press, New York, pp. 315–322. Winkworth, R.C., Robertson, A.W., Ehrendorfer, F., Lockhart, P.J., 1999. The importance of dispersal and recent speciation in the flora of New Zealand. J. Biogeogr. 26, 1323–1325.